Topics in Evolutionary Computation
Abstract
Autonomous robotic systems are expected to play a significant role in a wide range of areas including surveillance, deep space and undersea exploration and construction, urban search and recovery, mining, and hazardous waste cleanup. Systems that need to operate for extended periods of time out of range of human control should be adaptable to changing or unexpected conditions. This work examines some possible designs for such adaptive autonomous robotics systems, focusing on the adaptation to component failures in autonomous mobile robots. Adaptation is defined as the ability to continue to perform a task, perhaps at a degraded level, despite the loss of some of the robots original sensor and effector capabilities. The project addresses the problem of adaptation through an approach called Continuous Embedded Learning. Simulation and experimental results are reported.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 13, 2003
- Accession Number
- ADA417080
Entities
People
- John J. Grefenstette
Organizations
- George Mason University